National Poverty Center Working Paper Series
#08-10
July 2008
Lessons from Hurricane Katrina:
A Natural Experiment of the Effect of Residential Change on
Recidivism
David S. Kirk, University of Maryland
This paper is available online at the National Poverty Center Working Paper Series index at:
http://www.npc.umich.edu/publications/working_papers/
Any opinions, findings, conclusions, or recommendations expressed in this material are those of the author(s) and do
not necessarily reflect the view of the National Poverty Center or any sponsoring agency.
Lessons from Hurricane Katrina:
A Natural Experiment of the Effect of Residential Change on Recidivism
David S. Kirk
Department of Criminology and Criminal Justice
University of Maryland
* Direct correspondence to David Kirk, Department of Criminology and Criminal Justice,
University of Maryland, 2220 LeFrak Hall, College Park, MD 20742; email:
dkirk@crim.umd.edu. This work has been supported by a grant from the National Poverty Center
at the University of Michigan. I thank John Laub, Andy Papachristos, David Thacher, David
Weisburd, and Chris Wildeman for comments on an earlier draft of this manuscript as well as
participants of the National Poverty Center’s 2008 grantee workshop. I also thank Bill Sabol for
helpful advice on the conceptual and methodological design of the study. Any findings or
conclusions expressed are those solely of the author.
Lessons from Hurricane Katrina:
A Natural Experiment of the Effect of Residential Change on Recidivism
ABSTRACT
In August 2005, Hurricane Katrina ravaged the Louisiana Gulf Coast, damaging many of the
neighborhoods where ex-prisoners typically reside. Given the residential destruction resulting
from Hurricane Katrina, it is unclear whether the resulting geographic displacement of returning
prisoners has had any adverse, or even beneficial, impact on the likelihood of recidivism. It may
be the case that Hurricane Katrina induced turning points in the life course of former prisoners
by separating these individuals from the criminal peers and the criminal temptations that
contributed to their criminality in the past. Yet estimating the causal impact of residential
migration on the likelihood of recidivism is complicated by the issue of selection bias—the
possibility that some unmeasured characteristic of prisoners influences both where they live and
their criminal behavior, and may therefore account for any relation between residential change
and recidivism. In this study, I utilize a natural experiment as a means of addressing the selection
issue, and seek to establish whether the migration of ex-prisoners away from their former place
of residence will lead to lower levels of recidivism. Findings suggest that moving away from
former geographic areas substantially lowers an ex-prisoner’s likelihood of re-incarceration.
For the first time in my life, I didn’t have the feeling that I had to go to
Coxsackie, to Woodburn, and then to Sing Sing. I had the feeling now that
anything could happen, anything that I decided to do. It seemed a little bit crazy,
but I even had the feeling that if I wanted to become a doctor or something like
that, I could go on and do it. This was the first time in my life that I’d had that
kind of feeling, and getting out of Harlem was the first step toward that freedom.
Claude Brown (Manchild in the Promised Land, 1965:178)
INTRODUCTION
Hurricane Katrina initially developed on August 23rd, 2005 near the Bahamas, and made
its first landfall on the 25th near Ft. Lauderdale as a Category 1 hurricane. The storm then
moved into the Gulf of Mexico, and intensified, becoming a Category 5 hurricane by the
28th. At roughly 6:10am on August 29th, Katrina made its second landfall, near the
southern Louisiana town of Buras, and then moved inland towards New Orleans. Levee
walls on the Industrial Canal in New Orleans breached from the storm a couple hours
later, and additional levee breaches followed soon thereafter. The all-to-familiar
consequences are tragic. In Orleans Parish, 71.5 percent of housing units suffered some
damage following Hurricane Katrina, with 56 percent of housing units significantly
damaged (U.S. Department of Housing and Urban Development 2006). The extent of
housing unit destruction was similar in adjacent parishes. In both St. Bernard Parish and
Plaquemines Parish 80 percent of housing units were damaged, while 70 percent were
damaged in St. Tammany Parish and 53 percent were damaged in Jefferson Parish. More
than one million residents of New Orleans and the larger Gulf Coast region were
displaced by the hurricane (Liu et al. 2005).
The impacts of Hurricane Katrina on the Gulf Coast have been well documented,
yet one population group for which the implications of Katrina are still unclear is ex-
2
prisoners. As with the general population, the catastrophic property destruction and
associated residential displacement has repercussions for the population of offenders
exiting Louisiana prisons post-Katrina. This study aims to assess the repercussions of
Hurricane Katrina for released prisoners in Louisiana, and more so to examine the
implications of residential migration on the likelihood of recidivism. The tragedy of
Katrina presents an opportunity—a natural experiment—to assess the effects of
residential change on recidivism. If Claude Brown’s (1965) narrative tale of residential
change and desistance from crime generalizes to those former prisoners in Louisiana
induced to move because of Katrina, then we may find that Katrina actually led to
positive outcomes for this particular slice of the population. In other words, the lesson
from Katrina may be that residential change leads to a positive turning point in the lives
of former prisoners.
This study aims to answer the following research questions through an
examination of prisoner reentry1 in Louisiana: First, how has the geographic distribution
of prisoners released from Louisiana prisons changed following Hurricane Katrina?
Second, if there has been a geographic displacement of returning prisoners following the
hurricane, has this hurricane-induced migration had any negative or even beneficial
effects on the likelihood of recidivism?
Estimating the causal impact of residential migration on the likelihood of
recidivism is complicated by the issue of selection bias—i.e., the possibility that some
unmeasured characteristic of prisoners influences both where they live and their criminal
behavior, and may therefore account for any relation between residential change and
1
“Prisoner reentry” refers to the process of leaving prison and returning to the community (National
Research Council 2007).
3
recidivism. It may be the case that individuals with a high propensity towards criminal
offending and therefore a high likelihood of recidivism simply select into certain
geographic contexts, and that residential migration and the characteristics of these
contexts have little causal bearing on individuals’ behavior. If scholars of residential
migration and “neighborhood effects” are to make any kind of causal statements about
the impact of migration or place of residence on a given outcome measure, they must
address the issue of selection bias by separating out selection effects from any true effects
of residential change. In this study, I utilize a natural experiment as a means of
minimizing the potential for selection bias.
The paper proceeds as follows. As a starting point I situate the study by
describing the geographic context of prisoner reentry in the United States. I then outline
theoretically why geographic context is consequential for prisoner reentry and recidivism,
and therefore why a change of context may lead to a turning point in the life course of
crime. After that, I address two empirical objectives. I first describe the geographic
pattern of prisoner reentry in Louisiana before and after Hurricane Katrina by mapping
the addresses upon release for returning prisoners. I then estimate the causal effect of
hurricane-induced residential migration on the likelihood of recidivism among exprisoners.
THE GEOGRAPHIC CONTEXT OF PRISONER REENTRY
The Bureau of Justice Statistics reports that nearly 700,000 prisoners were released from
state and federal facilities in 2005 (Sabol and Harrison 2007), and estimates suggest that
up to half of these releasees have been in prison before (Langan and Levin 2002). By
4
some estimates, two-thirds of returning prisoners in the United States are re-arrested
within three years of prison release, and half are re-incarcerated (Langan and Levin
2002). These figures should not be separated from the geographic context in which
prisoners return. Ex-prisoners tend to be geographically concentrated in a relatively small
number of neighborhoods within the most resource deprived sections of metropolitan
areas, often returning to the same neighborhoods where they resided prior to
incarceration. For instance, research by the Urban Institute on reentry in Illinois reveals
that over half of prisoners released from Illinois prisons return to the city of Chicago, and
one-third of those returning to Chicago are concentrated in just six community areas (La
Vigne, Mamalian, et al. 2003). These six communities areas (Austin, East Garfield Park,
Englewood, Humboldt Park, North Lawndale, West Englewood) are among the most
economically and socially disadvantaged in the city. In Maryland, nearly 60 percent of
released prisoners return to the city of Baltimore, with 30 percent of those returning to
Baltimore concentrated in just six neighborhoods (La Vigne, Kachnowski, et al. 2003).
As with Chicago, released prisoners to Baltimore tend to reside in the most disadvantaged
sections of the city. Similar patterns of concentrated reentry into urban areas have also
been found in Georgia (La Vigne and Mamalian 2004), Ohio (La Vigne and Thomson
2003), and Texas (Watson et al. 2004).
As with prisoner reentry, prison admissions predominate from a select number of
neighborhoods within metropolitan areas. For instance, Lynch and Sabol (2001) find that
a mere 3 percent of the census block groups in Cuyahoga County in Ohio (which includes
Cleveland) account for over 20 percent of the state’s prison population, with an expected
350 to 700 ex-prisoners returning to those very same block groups each year following
5
prison release. Thus, prison admissions derive from select neighborhoods, prison releases
concentrate in those same neighborhoods, and most prisoners return “home” to the same
geographic areas where they resided prior to incarceration.
Given that many prisoners “return home” to their old neighborhoods, it is
important to develop a theoretical understanding for why these findings are
consequential. Insights from the life course of crime literature provide such a theoretical
foundation.
PLACE OF RESIDENCE, RESIDENTIAL MIGRATION, AND RECIDIVISM
The alarming rates of recidivism in the United States signify that there is much continuity
in criminal behavior over the life course. Sampson and Laub (1993) explain such
continuity by the absence of social controls (e.g., stable employment and a healthy
marriage), which is explained, in part, by the fact that a criminal record itself narrows the
opportunities for individuals to establish pro-social controls. Yet Sampson and Laub also
argue that continuity in behavior is not inevitable; change is possible, despite the
deleterious consequences that the mark of a criminal record has for future life events
(e.g., Pager 2003). Crucial to understanding the mechanisms behind changes in criminal
behavior is the notion of turning points, defined as “consequential shifts that redirect a
process” (Abbott 1997:101).
Hurricane Katrina may have induced turning points in the lives of former
prisoners for a number of different reasons. First, Katrina was a macro-level shock to the
Gulf Coast region. As life-course researchers have shown, historical shocks such as the
Great Depression and wars have profound consequences for the life course of affected
6
individuals (see, e.g., Elder 1974). A second, but no less important, implication of
Hurricane Katrina is residential change. Widespread property destruction in the Gulf
Coast region had an immense impact on residential relocation, some of which was
temporary and some permanent. In this study I focus on residential change as a turning
point in the life course of crime, and in the remainder of this section describe a number of
empirical studies and theoretical models which outline why residential change may be
consequential for diverting pre-existing criminal pathways.
In a study of the life course of crime from childhood through age 70, Laub and
Sampson (2003:149) contend that, “offenders desist [from crime] in response to
structurally induced turning points that serve as the catalyst for sustaining long-term
behavioral change.” Based on interviews with 70-year-old desisters, they find that
residential change is often a fundamental turning point that leads to desistance from
crime.2 Similarly, in an analysis of data from the Cambridge Study of Delinquent
Development, Osborn (1980) finds that delinquents who subsequently moved away from
London were significantly less likely to be reconvicted of a crime than delinquents who
stayed in London. Osborn investigates whether the lower level of reconviction among
movers is a function of changing attitudes, or a function of changing circumstances and
the elimination of criminogenic temptations, and finds support for the latter assertion.
Thus, in both the Laub and Sampson study and that of Osborn, authors find that a change
of residence allows individuals to separate from past situations and criminal peers, thus
eliminating some of the factors which contribute to the persistence of criminal behavior.
2
Desistance refers to the causal process that supports the termination of criminal offending (Laub and
Sampson 2001; 2003).
7
In the popular literature, perhaps the most illustrative example of the benefits of
separating from criminogenic environments is found in Claude Brown’s (1965)
semiautobiographical account, Manchild in the Promised Land. The lessons of crime and
delinquency began at an early age for Brown. As a 5-year-old, his neighborhood peers
taught him how to skip school and to avoid truant officers, and how to steal. During his
troubled youth, Brown committed countless crimes, used and sold a variety of drugs, was
expelled from school, asserted his dominance over both the Wiltwyck School for Boys
and the Warwick State Training School, and was even shot during a burglary. Yet after
an antisocial childhood, Brown was able to desist from crime while many of his peers
ultimately went to prison, became heroin addicts, or died (or some combination thereof).
One sure reason why Brown avoided such a fate is because he did not take to heroin as
did many of his peers and other residents of mid-century Harlem. Brown did not enjoy
his brief experimentation with the drug. Yet another compelling reason why he avoided a
life of crime is change of residence. As he entered adulthood, Brown made a conscious
decision to move away from Harlem and the peers, routines, and temptations of such a
familiar environment. He reflects (1965),
One thing began to scare me more than anything else about jail. This was the fact
that if I went to jail and got that sheet on me, any time I decided that I didn’t
want to go the crime way, that I wanted to do something that was straight, I’d
have a lot of trouble doing it behind being in jail. I didn’t want that sheet on me,
and I knew if I kept hanging around Harlem I was going to get busted. (P.177)
By moving to Greenwich Village, he was able to both physically and mentally separate
from many of the criminogenic influences in his life.
There are a number of potential mechanisms by which residential change may
lead to a turning point in the life course of crime, including a change in association with
criminal peers as well as a change in one’s routine activities. Association with criminal
8
peers may influence an individual’s criminality, in part, through a contagion process
whereby individuals learn the motivations and techniques for crime. Association with
criminal peers may also be consequential to crime because such associations provide
access to increased criminal opportunities. With respect to the former causal mechanism,
contagion models of crime generally posit that the likelihood of criminal and antisocial
behavior increases with exposure to others who engage in similar behavior. As Warr
(2002:3) convincingly demonstrates, “[C]riminal conduct is predominantly social
behavior. Most offenders are imbedded in a network of friends who also break the law,
and the single strongest predictor of criminal behavior known to criminologists is the
number of delinquent friends an individual has.”
Sutherland’s (1947) differential association theory is a prominent version of the
contagion model. Sutherland posits that criminal behavior is learned through interaction
in intimate social groups. This proposition helps explain one of the most robust findings
in the sociology of crime, namely that the bulk of crimes, particularly for youths, are
carried out in groups (Shaw and McKay 1931; Warr 2002). Sutherland also argues that
the process by which individuals learn criminal behavior is influenced by the broader
social organization in which they are embedded, including the organization of
neighborhoods. Put simply, in some neighborhoods individuals are more likely to be
exposed to other individuals who spread information about the techniques for committing
crimes and the motives and rationalizations for committing crimes.
If crime is in fact contagious, and is primarily a group phenomenon, then it
follows that removing individuals from their criminogenic social networks should reduce
their likelihood of engaging in criminal behavior. Research evidence supports such a
9
contention. For instance, Warr (1993) finds that exposure to delinquent peers declines
with age, and this explains the decline in crime with age. In a later work, Warr (1998)
dissects the reasons why marriage is such a powerful predictor of desistance from crime,
and argues that it is because marriage often disrupts peer relationships. He finds that
marriage reduces the likelihood that an individual will associate with delinquent peers,
and alters the amount of time individuals spend with delinquent and non-delinquent
peers. Moreover, his analysis reveals that the association between marriage and criminal
behavior disappears once controlling for peer relationship characteristics.
While Warr’s (1993; 1998) empirical findings may suggest a contagion
explanation for why peer associations are consequential, Laub and Sampson (2003) assert
that association with criminal peers is consequential because of the effect on an
individual’s routine activities and criminal opportunities. In other words, association with
delinquent peers does not necessarily mean that a given individual is learning the
motivations and techniques for committing crime; rather it may just mean that a given
individual is exposed to more opportunities for committing crime than s/he otherwise
would in the absence of criminal peers. For instance, through peers an individual may
have the opportunity to be an accomplice to a crime, or an individual may utilize a
criminal associate to fence stolen goods or to purchase drugs. Thus, separating an
individual from a criminal network via residential change may reduce her/his criminal
behavior by eliminating opportunities for engaging in crime. Regardless of whether peer
association is consequential because of contagion or because of the structuring of routine
activities and criminal opportunities, the implication of both explanations is that one way
to minimize criminal behavior is to separate individuals from their criminal peers.
10
Residential change may influence an ex-prisoner’s routine activities and therefore
criminal behavior beyond just the influence on peer association. For example, the cuereactivity paradigm in addiction research suggests that drug addicts are vulnerable to
relapse in the presence of familiar environmental stimuli (Wikler 1948; Carter and
Tiffany 1999). Through a process of classical conditioning, drug addicts come to
associate certain stimuli with the use of a drug. Thus, drug addicts are readily more likely
to relapse in those environments associated with prior drug use. In the presence of such
stimuli, drug addicts may encounter a physiological reaction to such cues including an
intense craving for drugs. Given that over half of state and federal prisoners have some
form of drug dependence or abuse as measured by the DSM-IV, and that an estimated
one-third of prisoners were under the influence of drugs when they committed the crime
for which they were imprisoned (Mumola and Karberg 2006:7), it reasons that residential
change may lower the likelihood of recidivism simply by separating drug addicts from
the “routine” activities which puts them in contact with familiar environmental stimuli.
There are numerous potential threats to internal validity, but one of particular
concern for the design of a study of residential migration or peer effects is the threat of
selection bias. Selection refers to the process of assigning individuals to conditions (i.e.,
treatment versus control groups). Selection bias may occur when an unobserved or
unmeasured characteristic of an individual influences both where s/he lives and the
outcome under study, and may therefore account for any relation between residential
change and the outcome. Similarly, selection bias may occur when an unmeasured
characteristic of an individual influences her/his selection into a peer group and the
outcome under study, and may therefore account for any relation between peer effects
11
and the outcome. The theoretical implication derived from the life-course literature is that
to lessen the probability of recidivism and increase the likelihood of desistance, it is
beneficial to separate ex-prisoners from their criminal past and their peers. Yet, sorting
out the causal consequences of residential migration and peer effects requires a research
design that explicitly addresses the issue of selection bias.
In this study I attempt to answer the first part of a two part causal story. I seek to
uncover whether separating individuals from their former criminogenic environment
reduces their likelihood of recidivism, and do so through a research design which
expressly accounts for the potential of selection bias. If residential migration does reduce
the likelihood of recidivism, the second part of the causal story is to investigate why. Two
potential answers are changes to routine activities and peer associations. I save the
investigation of why for future analyses.
DATA AND RESEARCH DESIGN
This study exploits the benefits of a natural experiment in order to characterize the
geographic distribution of prisoner reentry in Louisiana and assess the repercussions of
geographic displacement from Hurricane Katrina on the probability of re-incarceration.
The analytic sample is drawn from prisoners released from Louisiana correctional
facilities. Because my interest is in the displacement due to Hurricane Katrina, I restrict
analyses to those prisoners who resided in affected metropolitan areas prior to
incarceration. Accordingly, the analytic sample only includes ex-prisoners who were
committed to prison from Orleans Parish and the four parishes adjacent to Orleans
(Jefferson, Plaquemines, St. Bernard, and St. Tammany). These areas sustained
12
substantial housing damage in the days following Hurricane Katrina. Thus, for those
prisoners released since Hurricane Katrina, arguably their residential choices are
significantly different than if they had been released prior to the hurricane, resulting in
some measure of geographic displacement. Finally, I also restrict the sample to exclude
sex offenders. Given the nature of their offense, sex offenders face a number of
constraints on their residency choices upon release from prison. Because of this, I opt to
exclude these individuals from the study.
For the purposes of analyses, I construct three cohorts of prison releasees, two of
which were released from prison prior to Hurricane Katrina and one released afterwards.
The first cohort is comprised of first-releases from a Louisiana prison to parole
supervision anytime from September 2001 to February 2002 (hereafter called the 20012002 cohort).3 By first-releases, I refer to individuals who were released from their first
stint of incarceration in a Louisiana prison. The second cohort consists of first-releases
onto parole supervision between September 2003 and February 2004 (the 2003-2004
cohort). The third cohort consists of first-releases onto parole supervision between
September 2005 and February 2006 (the post-Katrina cohort). The selection of years for
the 2003-2004 cohort is designed to allow for a one-year observation period for reincarceration, which still occurs prior to events associated with Hurricane Katrina.
3
I use the term “parole supervision” to refer to all release mechanisms from incarceration that result in a
term of supervision for the returning prisoner. Releases to supervision by the Louisiana Department of
Public Safety & Corrections are generally made through parole board action or through the diminution of
the sentence via good time credit. “Parole supervision” encompasses both types of releases. Roughly 90
percent of prisoners released each year from Louisiana prisons are released onto parole supervision (in
contrast to unconditional releases, which do not require post-incarceration supervision). Prisoners to be
released onto parole supervision are required to provide the Division of Probation and Parole a residence
plan prior to release, which includes the parish and address where they will reside. Prisoners in Louisiana
are not required to return to the same parish where they were convicted. To distinguish my analytic sample
of paroled releases from unconditional releases, throughout the analysis I will utilize the term “parolee.”
13
Sample sizes equal 1,174, 1,214, and 887 for the 2001-2002, 2003-2004, and 2005-2006
cohorts, respectively.
For a true randomized experiment, it would be appropriate to assume that the
three release cohorts are balanced (i.e., equivalent) with respect to observable and
unobservable characteristics. Without random assignment, it is not necessarily the case
that the three cohorts are equivalent. My rationale for only sampling parolees released
from their first term of imprisonment stems from the fact that parolees as a whole are a
very heterogeneous group with widely varying rates of recidivism (National Research
Council 2007). More specifically, parolees released from their first term of imprisonment
have, on average, substantially lower rates of recidivism than parolees with multiple prior
incarcerations. Rosenfeld and colleagues (2005) find that first-releases from incarceration
accumulate between 18 and 25 percent fewer subsequent arrests during the three years
immediately following incarceration relative to those ex-prisoners who had at least one
prior incarceration. Given that I have limited information on criminal history, I restrict
my sample to first-releases to ensure equivalence across cohorts on criminal history (i.e.,
each cohort, on average, is characterized by similar incarceration histories). Another
consequence of restricting my sample to first-releases is that later cohorts will not contain
any of the same parolees as earlier cohorts.
Data utilized in this study are of three varieties: a) individual-level data on
parolees from the Louisiana Department of Public Safety & Corrections (DPS&C) and
the Division of Probation and Parole (DPP)4, b) zip code and parish-level characteristics
from the U.S. Department of Housing and Urban Development, the Louisiana
4
DPS&C data is maintained in the Corrections and Justice Unified Network (CAJUN) database, and DPP
data is maintained in the Case Management System.
14
Department of Labor, and ESRI, and c) Louisiana criminal justice system data from the
Supreme Court of Louisiana, DPS&C, DPP, and the Uniform Crime Reports.
Given that social and economic conditions in Louisiana changed drastically
immediately following Hurricane Katrina, it is necessary to control for such temporal
changes in order to make causal statements about the relation between residential
migration and recidivism. As mentioned in the previous section, the macro-level shock
associated with Katrina may induce turning points in the life course of crime for postKatrina parolees, just as residential migration may also induce turning points. Controlling
for macro-level social and economic changes to the region allows me to account for such
a macro-level shock, and therefore isolate the effect of residential change on recidivism.
As but one example of temporal change to the region, estimates a little over one year
after Hurricane Katrina placed the population of Orleans Parish at just 191,139 residents
(as of October 2006; Louisiana Public Health Institute 2006). Estimates two years after
the storm placed the population at 264,969 residents (ESRI 2007). In contrast, the
estimated population of Orleans Parish just prior to Hurricane Katrina was 437,186
residents (Brookings Institution 2006).
ESRI provides yearly demographic estimates of the U.S. population at both the
county/parish level and zip code level. I utilize these demographic estimates to produce
measures of racial segregation and household income. To derive post-Katrina estimates,
ESRI (2006a; 2006b) augmented their standard methodology used to compute intercensal
demographic estimates by incorporating data from the Federal Emergency Management
Agency (FEMA) on property damage and applications for assistance, from the United
States Postal Service National Change of Address file, and from the American Red Cross
15
on estimated housing unit damage.5 ESRI estimates for 2006 reflect data current as of
January and February of 2006 (ESRI 2006b).
Similar to the impact on social and economic conditions, the implications of
Hurricane Katrina for temporal changes in the criminal justice system in Louisiana are
many (for detailed discussions, see Garrett and Tetlow 2006; Roman, Irazola, and
Osborne 2007). Most obviously, many justice officials evacuated from New Orleans in
the wake of the storm, and some did not return. The Orleans Parish District Attorney’s
office flooded, and office staff were forced to work from temporary locations following
the storm. The Orleans Parish courthouse also flooded. Makeshift courts were setup in
prisons and other facilities, but the result was ultimately a delay in legal proceedings in
the Orleans Parish Criminal District Court. Given the impact of Katrina on the criminal
justice system, particularly in Orleans and adjacent parishes, it is vital to account for
temporal variation in the operation of the justice system in order to draw any kind of
causal inference about the effect of residential change on recidivism. Thus, I include
control variables in analyses related to parole practices, court operations, and the
probability of arrest given the commission of a crime.
INDIVIDUAL-LEVEL VARIABLES
RE-INCARCERATION. Re-incarceration refers to whether a given parolee returned to
a Louisiana prison for a new criminal conviction or a parole violation within one year of
prison release. This is the primary measure of recidivism used in the study.
5
Description of the ESRI methodology can be found in the cited references, and from the following
website: http://www.esri.com/data/community_data/demographic/methodology.html. Data used in this
study are available for purchase from ESRI, and are also published in annual editions of the Community
Sourcebook of County Demographics and the Community Sourcebook of ZIP Code Demographics.
16
RE-INCARCERATED OR DETAINED. While I rely upon Louisiana data in this study, it
may be true that a proportion of the release cohorts from Louisiana will be committed to
prison in another state or in the federal system. However, since the three cohorts include
only those prisoners released to parole supervision, the commitment of parolees to other
state correctional systems may be limited, given that parolees are required to remain in
Louisiana and make periodic visits with their parole officer. Information on nonLouisiana incarcerations, as well as incarcerations in a local facility (as opposed to state),
is captured in the DPP case management system in a field for supervision level. A
parolee’s supervision will be marked as “detained” if that parolee is confined in another
jurisdiction.6 In the interest of determining whether my inferences are sensitive to the
operationalization of recidivism, I re-estimate my statistical model with a second measure
of recidivism. This second, more inclusive measure combines the RE-INCARCERATION
measure described previously with information on other detentions. In other words, this
secondary measure of recidivism indicates whether a given parolee was re-incarcerated in
a Louisiana prison within one year, or was detained in another jurisdiction.
Different Parish as Conviction. This is a binary variable indicating whether the
parolee moved to a different parish following incarceration relative to where s/he was
originally convicted. In contrast to some other states, prisoners in Louisiana are not
required to return to the same parish where they were convicted or last resided. I
geocoded the address of residence for a given parolee immediately following her/his
6
Information on other jurisdictional detention is obtained if the DPP directly transfers a given parolee to
the other jurisdiction, or through a process of tracking absconders. In the event that a parolee absconds and
misses a periodic visit with her/his parole officer, the officer may decide to issue a warrant for the parolee
through the FBI’s National Crime Information Center (NCIC) and through local law enforcement. To track
absconders, DPP parole officers also search the National Law Enforcement Telecommunications System
(NLETS) as well as local warrant systems to identify the location of absconders. If absconders are detained
in a local or non-Louisiana facility, information in the DPP data system is updated accordingly.
17
release from prison, and compared this to the parish of conviction. This variable equals
zero for those parolees who returned to the same parish as where s/he was convicted, and
one for parolees who migrated to a different parish.
Post-Katrina Release. This is a binary variable indicating whether the parolee
was released from prison following Hurricane Katrina. As I describe to follow in the
section on Analytic Strategy, this variable is used as an instrument in analyses.
The study includes five additional individual-level measures as correlates of
recidivism, all derived from prisoner data in the DPS&C CAJUN database system: race,
gender, age at time of release, marital status, and time served.7 Black parolees comprise
73.3 percent of the sample, with whites making up 26.5 percent. Other races make up 0.2
percent of the sample. I use a binary indicator (Black) in analyses to compare black
versus white-other race categories. Male is a binary indicator of gender, with female as
the reference category. Males comprise 87 percent of the sample. Married is a binary
variable indicating the marital status of the parolee at the time of her/his release (not
married equals zero). Eleven percent of the sample was married at the time of release.
Time served refers to the amount of time a parolee served in prison (in years or fraction
thereof) until release. Controlling for time served is crucial in order to account for any
differences between cohorts in the average severity of prior offending. In other words, if
in the post-Katrina period the likelihood of a prisoner being released from custody once
s/he is eligible declined relative to the pre-Katrina period, accounting for time served will
capture differences across release cohorts in prior offending severity.
7
Time served is highly associated with the offense of conviction (e.g., prisoners convicted of violent
offenses serve more time in Louisiana relative to other offenses). Thus, in the interest of minimizing
collinearity, I use time served as a control in analyses but not indicators of offense of conviction.
18
CONTEXTUAL-LEVEL VARIABLES8
Dissimilarity. Dissimilarity (D) is a measure of the evenness of population
distribution (Duncan and Duncan 1955). In the specific case here, it is a measure of
segregation between blacks and whites, reflecting their relative distributions across zip
codes within each parish. It can range in value from 0, indicating complete integration, to
100, indicating complete segregation. For the 2001-2002 cohort, I use an estimate of D
for a given parolee’s parish of residence calculated from the 2002 ESRI data, while for
the 2003-2004 and 2005-2006 cohorts I use estimates of D derived from the 2004 and
2006 data, respectively.
Average Household Income (Sqrt). This is a measure of the average household
income (in 2000 adjusted dollars) in the zip code in which a given parolee resides. For
the 2001-2002 cohort, I use the 2002 ESRI household income estimates, and for the
2003-2004 and 2005-2006 cohorts I use respective estimates from the 2004 and 2006
data. To reduce the skewness of the measure, I utilize the square root of income in the
statistical models estimated to follow.
Unemployment Rate. This is a measure of the parish unemployment rate in a
parolee’s parish of residence in the quarter during which the prisoner was released from
prison. Data derive from the Louisiana Department of Labor.
8
Research has shown that the level of aggregation (e.g., census block, tract, zip code, city, metropolitan
area, and county) used in studies of contextual effects influences inferences derived from the association
between a given contextual characteristic and a dependent variable (Hipp 2007). For the contextual
measures described to follow, I utilize zip code and parish levels of aggregation. I do so partially because
measures such as unemployment and wages are reflective of metropolitan-wide economic conditions, and
also because monthly, quarterly, and yearly time series data is not available at a lower level of aggregation
(e.g., census tracts).
19
Average Weekly Wage. This is a measure of the average weekly wage (in 2000
adjusted dollars) in the parish where a given parolee resides for the quarter during which
s/he was released from prison. Data derive from the Louisiana Department of Labor.
Fair Market Rent. This is a measure of the average fair market rent in the parish
where a given parolee resides. Data derive from the U.S. Department of Housing and
Urban Development (HUD). For the 2001-2002 cohort, I use the fair market rent in 2002
for a given parolee’s parish of residence, while for the 2003-2004 and 2005-2006 cohorts
I utilize rent averages from 2004 and 2006, respectively. All figures are adjusted to 2000
dollars.
CRIMINAL JUSTICE SYSTEM VARIABLES
Average Parole Contacts. This measure derives from the DPS&C Quarterly
Statistical Performance Report, which provides data on the operations of the correctional
and parole systems, including information on the average number of interpersonal
contacts made per parole officer in a given parole district. Research suggests that there is
an inverse relationship between parole officer caseload and recidivism, because smaller
caseloads lead to more contacts with parolees and more scrutiny (Turner and Petersilia
1992). Thus, it is vital to control for temporal variations in the average parole contacts
(i.e., the total contacts a parole officer has across her/his entire caseload), as well as
between parole district variation. The measure used in analyses is the average number of
contacts made across parole officers in the parolee’s parole district during the quarter in
which s/he was released from prison. Contacts include those which occurred within the
parole office as well as field contacts to the parolee’s residence or workplace.
20
Judge Caseloads. This measure derives from the Louisiana Supreme Court’s
annual report, which provides data on the operations of the Louisiana court system. Each
year, the Supreme Court reports the number of cases filed in each District court as well as
the number of judges. There is a vast body of research which examines the relation
between judge caseloads and criminal case dispositions (e.g., Dixon 1995; Feeley 1979;
Heumann 1975). One key issue in this regard is whether large caseloads pressure courts
towards the use of plea bargaining and result in lenient sentencing practices because of
the need to process large numbers of defendants (e.g., Eisenstein, Flemming, and
Nardulli 1988; Nardulli 1979). Recent research (see, e.g., Johnson 2005; Ulmer and
Johnson 2004) reveals that judge caseloads are inversely related to the likelihood of
incarceration and the severity of criminal sentences. Thus, I include a control for judge
caseloads given that such caseloads likely influence whether a convicted offender is
sentenced to a term of imprisonment or some other sanction, such as probation or parole.
For the 2001-2002 cohort, I use the number of cases per judge in 2002 in a given
parolee’s parish of residence, while for the 2003-2004 and 2005-2006 cohorts, I use
caseload figures from 2004 and 2006, respectively.
UCR Arrests per Crime. This measure derives from the FBI’s Uniform Crime
Reports, and is used to control for the temporal and geographic variation in the likelihood
of getting arrested for a Part I offense (i.e., murder, rape, robbery, aggravated assault,
burglary, larceny, motor vehicle theft, and arson) given the commission of such a crime.
For the 2001-2002 cohort, I construct this measure using the ratio of Part I arrests per
reported Part I crimes in 2001 for a given parolee’s parish of residence, while for the
2003-2004 and 2005-2006 cohorts I use the 2003 and 2005 ratios, respectively.
21
ANALYTIC STRATEGY
Analyses follow two paths to coincide with the objectives of the study. In the first set of
analyses, I test two hypotheses. I expect that prisoners released soon after Katrina faced
severely constrained opportunities to reside in New Orleans, resulting in far fewer returns
to the New Orleans metropolitan area than in the pre-Katrina period and fewer returns to
former parishes. This expectation leads to Hypothesis 1a:
Hypothesis 1a. Because of hurricane-related destruction, significantly fewer
parolees moved to New Orleans following incarceration during the time period
immediately following Hurricane Katrina relative to the pre-Katrina period, and
significantly fewer returned to the parish they inhabited prior to incarceration.
However, I also expect that the structural barriers that lead to a concentration of
prisoner reentry into the most impoverished urban neighborhoods will result in exprisoners migrating to other disadvantaged urban areas besides those found in New
Orleans. In addition, the residential patterns of prisoner reentry are likely affected by the
increased competition for housing resulting from the forced migration of Louisiana Gulf
Coast residents to geographic areas unaffected by Hurricane Katrina. These observations
lead to Hypothesis 1b:
Hypothesis 1b. The types of geographic areas where parolees typically reside, in
terms of poverty and economic conditions, are similar or even more disadvantaged for
the post-Katrina period relative to the pre-Katrina period.
To examine these hypotheses, I geocode and map the post-release addresses for
the three separate cohorts of parolees, and provide illustrations of how the geographic
distribution of prisoner reentry changed post-Katrina. To address Hypothesis 1b
22
specifically, I compare the three parolee cohorts with respect to the five measures of
social and economic context described previously.
With regards to my second research question concerning the impact of residential
change on recidivism, I expect that separating parolees from their former residential
environments will be beneficial with respect to desisting from crime. In other words, I
propose that the induced migration due to Hurricane Katrina serves some benefit in that it
allows for a separation between parolees and their criminal past, thus reducing the
likelihood of re-incarceration (even if the types of geographic areas where parolees reside
become more disadvantaged). These propositions yield Hypotheses 2a and 2b:
Hypothesis 2a. The proportion of ex-prisoners re-incarcerated for a new
conviction or parole violation is significantly lower in the post-Katrina period.
Hypothesis 2b: The likelihood of re-incarceration is lower when parolees reside
in a geographic area different from where they resided prior to incarceration.
Hypotheses 2a and 2b are related, but do address slightly different questions.
Hypothesis 2a tests the “Katrina Effect”—i.e., the difference in the likelihood of reincarceration for those prisoners released post-Katrina relative to those released preKatrina. In contrast, Hypothesis 2b is focused specifically on the impacts of residential
migration on recidivism. The distinction lies in the fact that some parolees post-Katrina
did return to the parish where they resided prior to incarceration. In other words, while
Katrina is consequential to residential migration following imprisonment, not everyone
moved to a different parish.
I employ two different modeling strategies to address these hypotheses: an
intention-to-treat analysis (ITT) to address Hypothesis 2a and an instrumental variables
23
analysis (IV) to address Hypothesis 2b. For the ITT analysis, I provide an estimate of the
difference between the post-hurricane (the treatment group) and two pre-hurricane
cohorts (the control group) on the likelihood of re-incarceration. In the language of
counterfactual reasoning, my interest is in estimating the recidivism outcome for a given
parolee if s/he was released from prison for the first time following Hurricane Katrina
versus the counterfactual condition of being released from prison for the first time prior
to Katrina. For each specific parolee, however, we can only observe one outcome. Thus,
it is not possible to directly calculate the causal effect of Hurricane Katrina for a specific
prisoner. Given this restriction, one alternative is to use an exogenous source of variation
to assign comparable individuals to control and treatment groups, and then estimate the
treatment effect. In this case, the control individuals serve as the counterfactual.
As noted, with a true randomized experiment, it is assumed that treatment and
control groups are balanced (i.e., equivalent) with respect to observable and unobservable
characteristics, aside from random variation. In the present study, I assume that the type
of prisoner released post-Katrina will be similar to those prisoners released pre-Katrina.
Still, without random assignment, it is not necessarily the case that treatment and control
groups are identical. Thus, it is beneficial to include statistical controls for key parolee
characteristics and other relevant factors in statistical models in order to account for any
differences. In equation form, re-incarceration is estimated as a function of the release
cohort (i.e., the ITT effect, Zi) and a set of controls:
Yi = Ziθ1 + Xiβ + εi
(1)
where
Yi is a dichotomous variable indicating whether parolee i was re-incarcerated within
one year of release;
24
Xi is a vector of control variables used to account for any observed differences
between the pre- and post-hurricane cohorts;
Zi is a binary indicator of treatment assignment, which in this case is the release
cohort (i.e., the Post-Katrina Release binary variable).
Because the treatment and control groups are defined by the time period of prison
release (i.e., before or after the hurricane), in essence the ITT estimate provides
inferences as to whether there are any differences in re-incarceration between the two
groups. While informative, the ITT estimates provide no information on causal
mechanisms. In other words, it is not possible to draw conclusions as to why the preversus post-Katrina cohorts differ on recidivism. However, combining the natural
experiment method used to estimate the ITT with an IV approach allows for a greater
examination of the mechanisms by which the geographic displacement of prisoner
reentry influences recidivism. Thus, to test Hypothesis 2b I compare the likelihood of reincarceration with one-year of release from prison for parolees who resided in the same
parish upon release as where they were originally convicted versus parolees who moved
to a different parish. Here, the causal mechanism is residential migration.
In addition to the exploration of causal mechanisms, combining a natural
experiment with an IV approach allows me to further control for selection bias. With a
natural experiment alone, a simple comparison of the treatment and control groups may
be biased by unknown or unmeasured factors which influence residential migration
(besides whether the parolee was released pre- or post-Katrina). Combining the natural
experiment with an IV approach can provide an unbiased estimate of the effect of moving
to a different parish on re-incarceration (Angrist, Imbens, and Rubin 1996). With an IV
approach, a variable (or variables) that is unrelated to the outcome variable is used as a
25
predictor (i.e., instrument) of the key explanatory variable (i.e., the treatment), and then
the outcome variable is regressed on the predicted treatment measure. Conceptually, this
approach removes the spurious correlation between the explanatory variable and
unobserved characteristics, in this case unobservable characteristics of prisoners. The key
criticism of this approach is that the assumption about the lack of relation between the
instrument and the outcome variable may be problematic. However, the use of an
instrument derived from a natural experiment obviates this issue. In other words, we can
have more confidence that the instrument and outcome variable are unrelated if that
instrument derives from a random force of nature like a hurricane (for a discussion, see
Angrist and Krueger 2001). This assumption is known as the exclusion restriction—i.e.,
cov(Zi,ui ) = 0—and is not directly verifiable (Angrist et al. 1996). The validity of
inferences from an IV analysis depends upon the appropriateness of this assumption. In
the Discussion section following the presentation of results, I address potential violations
of the exclusion restriction.
To test Hypothesis 2b, the treatment condition is whether, upon release from
prison, a parolee migrated to a parish different from where s/he was originally convicted.
Thus, the control group is comprised of prisoners who returned to the same parish upon
release, while the treatment group consists of prisoners who migrated to a different parish
upon release. An IV remedies the issue of omitted variables (e.g., the fact that I lack
complete information for why a given parolee selects one parish of residence versus
another) by using only that portion of the variability in the treatment variable that is
uncorrelated with omitted variables to estimate the causal relation between the treatment
and outcome. As Angrist and Krueger note (2001:77), “instrumental variables provide an
26
estimate for a specific group—namely, people whose behavior can be manipulated by the
instrument.” In the present context, Hurricane Katrina affected the residential behavior of
those parolees who could not or would not have moved away from the New Orleans
metropolitan area following incarceration, but did not affect those parolees who would
have moved even in the absence of the hurricane.9 In short, use of an IV approach allows
me to compute an estimate of the effect of migrating to a different parish for those
parolees who otherwise would have moved back to the same parish instead had Hurricane
Katrina not occurred (this is termed the Local Average Treatment Effect, or LATE). I do
not provide an estimate of the effect of moving to a different parish on re-incarceration
for those parolees who would have moved regardless of the hurricane.
The two structural equations specified to follow outline the two-stage estimation
process with the IV technique. The first equation models the key explanatory variable Si,
a binary measure indicating whether parolee i resides in a different parish upon release
relative to where s/he was convicted prior to incarceration, as a function of an
instrumental variable Zi and a vector of control variables Xi. The instrument Zi is a binary
indicator of the release cohort (post-Katrina versus otherwise). It is assumed that where a
parolee resides depends, in part, upon whether s/he was released from prison before or
after Hurricane Katrina.
Si = Ziθ1 + Xiπ + ξit
9
(2)
As an example, Angrist (1990) uses Vietnam-era draft lotteries as an instrument to assess the effect of
military service on future earnings. Some individuals who joined the military did so voluntarily, and some
did so because of the draft. His IV estimates of the effects of military service on future earnings are
applicable for draftees but not volunteers. In other words, he estimates the effects of military service on the
earnings of draftees, yet does not capture the effect of military service on earnings for volunteers.
27
The second stage of the two-stage estimation process models the dependent variable (Yi),
a dichotomous variable indicating whether parolee i was re-incarcerated within one year
of release, as a function of the predicted Si from Eq.2 and a vector of control variables Xi:
Yi = α S i + X i β + u i
(3)
The coefficient α is the key parameter of interest, and will be used to draw inferences as
to whether migrating to a new parish upon release influences a parolee’s likelihood of
recidivism.10
RESULTS
HYPOTHESES 1A AND 1B
To examine Hypotheses 1a and 1b, I present a series of tables as well as a descriptive
figure to compare and contrast the geographic patterns of prisoner reentry pre- and postKatrina. Figure 1 provides a snapshot of the geographic redistribution of parolees postKatrina. More specifically, this figure reveals in which parish parolees reside
immediately upon their exit from prison. The release figures from 2001-2002 and 20032004 reveal that roughly 45 percent of prisoners convicted in the 5-parish area (i.e.,
Orleans, Jefferson, Plaquemines, St. Bernard, and St. Tammany) subsequently returned to
New Orleans. Post-Katrina, this number drops to 20 percent. In the post-Katrina period,
proportionally more parolees opted to reside in Baton Rouge (2 percent before Katrina
versus 9 percent after), and many other parolees dispersed throughout the state. This
figure demonstrates that the proportion of newly released prisoners (who were originally
committed from the New Orleans metropolitan area) who returned to the 5-parish area
10
I estimate equations (2) and (3) in STATA using the ivprobit function. Model standard errors are adjusted
to account for the clustering of parolees within parishes.
28
upon release declined drastically immediately following Katrina. This finding provides
support for Hypothesis 1a. Moreover, recall that one benefit of using an instrument
derived from a natural experiment is to assure that said instrument is correlated with the
treatment condition (i.e., migrating) but otherwise unrelated to the outcome variable (i.e.,
re-incarceration). Results presented in Figure 1 suggest that the time period of release
(pre versus post-Katrina) is highly correlated with the treatment condition.11
[FIGURE 1 ABOUT HERE]
Table 1 contrasts the place of residence across the three cohorts through a crosstabulation of the proportion of members of each cohort who moved to the same parish
following incarceration as where they were originally convicted relative to the proportion
who migrated to a different parish. Results reveal that prior to Hurricane Katrina, roughly
three-quarters of parolees returned to their parish of conviction upon release from prison.
Post-Katrina, this distribution changed significantly (Chi-Square = 127.405, p < 0.001),
with 53.2 percent of parolees returning to the same parish and 46.8 percent migrating to a
different one. This finding provides additional support for Hypothesis 1a.
[TABLE 1 ABOUT HERE]
To examine Hypothesis 1b, Table 2 presents a descriptive summary of the
geographic areas where the respective cohorts reside. This table displays the percentages
of parolees within each cohort which fall within a given category along the distributions
of the five measures I utilize to describe social and economic context.
11
A key assumption of the IV framework is that the covariance between the treatment and instrument
differs from zero: cov(Si, Zi ) ≠ 0. Results from Figure 1 support this assumption. The substantial
correlation between the instrument and the treatment is also confirmed via the ivprobit estimation—
equations (2) and (3) are estimated jointly, and the correlation between Post-Katrina Release and Different
Parish as Conviction is positive and highly significant (p < 0.001).
29
The impacts of Hurricane Katrina were felt proportionally more in those New
Orleans neighborhoods marked by poverty and high proportions of black population, and
it is in these types of urban neighborhoods where prisoners disproportionately reside
upon release from prison (e.g., Clear, Waring, and Scully 2005). For instance, in the city
of New Orleans, 29.2 percent of the population in damaged areas had incomes below the
poverty line in comparison to 24.7 percent of the population in unaffected areas (Logan
2006). Furthermore, the population of damaged areas was 75 percent black compared to
46.2 percent in unaffected areas. Because of the disproportional impact of the storm,
demographic shifts following Hurricane Katrina have altered the evenness of the
population distribution in the city of New Orleans and the wider metropolitan area.
Population estimates from the ESRI data as well as enrollment figures from New Orleans
public schools (Brookings Institution and Greater New Orleans Community Data Center
2007) reveal that black residents have been relatively less likely to return to New Orleans
post-Katrina than white residents. Specifically, data reveal that prior to the hurricane,
black residents comprised roughly 68 percent of Orleans Parish residents and whites
made up 27 percent. Post-Katrina, the share of black population has declined to 44
percent and the share of white population has increased to 50 percent. Estimates also
suggest that Latino groups have migrated to the metropolitan area in substantial numbers
in the post-Katrina period. Given that a sizable portion of parolees post-Katrina still
reside in the metropolitan area, the demographic shifts in New Orleans impact the
likelihood that parolees will reside in an area characterized by segregation. Findings
presented in Table 2 support this assertion. Post-Katrina parolees are significantly more
30
likely to reside in an area characterized by segregation and less likely to reside in an
integrated area relative to their pre-Katrina counterparts.
[TABLE 2 ABOUT HERE]
Trends in the unemployment rate and weekly wages combine to produce
offsetting economic consequences for returning prisoners. Reports suggest that the New
Orleans metropolitan area lost roughly 4,000 firms in the year following Katrina,
equating to 15 percent of the regional employer base (Brookings Institution and Greater
New Orleans Community Data Center 2007). At the same time, the labor force size
plummeted with the evacuation of the metropolitan area; yet, it has been steadily
increasing since the first anniversary of the storm. Unemployment rates in the New
Orleans metropolitan area initially spiked following the storm, reaching 15.8 percent in
November 2005. Yet owing to the decline in the labor force and the availability of jobs to
rebuild the metropolitan region, unemployment declined to as low as 3.8 percent by
December 2006, well below the national average and pre-Katrina levels. Spurred by a
shortage of workers in key industries such as construction and hospitality, the average
weekly wage in the New Orleans metropolitan area increased significantly following
Katrina. For instance, construction wages increased by 41 percent one year after Katrina,
and wages for professional and technical services jumped 53 percent (Brookings
Institution and Greater New Orleans Community Data Center 2007). What have these
trends meant for prisoners released soon after Hurricane Katrina? Data presented in Table
2 reveal that post-Katrina parolees resided in areas with significantly higher wages than
those individuals released from prison pre-Katrina. At the same time, a significantly
higher unemployment rate characterized the post-Katrina context for parolees. The
31
average household income (by zip code) remained similar pre- and post-Katrina, which is
partially a function of the offsetting effects of unemployment and wage increases.
With fair market rents, temporal patterns of prisoner reentry reflect the fact that
the supply of rental housing in the New Orleans metropolitan area declined significantly
following Katrina due to property destruction, while at the same time the demand for
rental units from displaced residents significantly increased (Brookings Institution and
Greater New Orleans Community Data Center 2007). HUD data reveal that the average
fair market rent in Orleans Parish rose a staggering 39 percent from 2005 to 2006, from
$676 for a two-bedroom apartment to $940. Rents also rose in other geographic areas
where parolees concentrated post-Katrina. For instance, rents climbed 18.4 percent in
Baton Rouge, from $608 to $720 for a two-bedroom apartment. The consequence is that
post-Katrina parolees tended to reside in geographic areas characterized by significantly
higher rents than their counterparts released prior to Katrina.
Taken as a whole, findings presented in Tables 1 and 2 as well as Figure 1 suggest
that prisoners released post-Katrina may be residing in different geographic areas, but
these areas are not any more advantaged relative to the types of geographic contexts
where parolees concentrated pre-Katrina. If anything, the socio-economic context of
prisoner reentry became slightly more disadvantaged following the hurricane. Such
findings provide support for Hypothesis 1b.
HYPOTHESES 2A AND 2B
Figure 2 examines whether parolees who returned to the same parish as where they were
originally convicted were any more or less likely to recidivate relative to parolees who
32
migrated to different parishes. While re-incarceration has been declining in more recent
years, it is also true that for all three cohorts those parolees who returned to the same
parish as where they were originally convicted were more likely to be back in prison
within one year.
[FIGURE 2 ABOUT HERE]
Table 3 demonstrates the effect of Hurricane Katrina on re-incarceration (i.e., the
ITT results). As expected, males are more likely to be re-incarcerated than females.
Those parolees who are married are less likely to be re-incarcerated. Thus, as
documented in numerous studies (e.g., Laub and Sampson 2003; Warr 1998), marriage
has a controlling influence on criminal behavior, and is a significant contributor to
desistance from crime. Time served in prison is negatively related to re-incarceration.
With respect to the contextual-level and criminal justice system covariates, only fair
market rents are significantly associated with re-incarceration. In this case, higher rents
correspond to a greater likelihood of re-incarceration, which may indicate that parolees
are turning to crime in order to keep up with rising costs of housing or that higher rents
signal more opportunities for crime.
[TABLE 3 ABOUT HERE]
The key parameter of interest in Table 3 is that for Post-Katrina Release. Results
reveal that parolees released following Katrina are significantly and substantially less
likely to be re-incarcerated within one year of release. Translating these findings to a
marginal effect, the probability of re-incarceration within one year of prison release is
0.09 lower for parolees released post-Katrina relative to parolees in the 2001-2002 cohort
and the 2003-2004 cohort (net of individual, contextual, and criminal justice system
33
correlates). The predicted probability of re-incarceration for male parolees released prior
to Katrina is 0.204, while the predicted probability for males released after the hurricane
is 0.112. These findings provide support for Hypothesis 2a.
Results thus far reveal that there has been a geographic redistribution of prisoner
reentry post-Katrina and that recidivism declined. Whether these two processes are
causally related is an empirical question, which I will address through an IV analysis.
Before moving onto the IV results, it is necessary to address the topic of treatment
noncompliance. Recall that in the present case the treatment Si is represented by a given
parolee residing in a different parish upon release relative to where s/he was convicted
prior to incarceration. Perfect treatment compliance would represent the situation where
all parolees released post-Katrina (Z = 1) moved to a different parish (S = 1), and all
parolees released pre-Katrina (Z = 0) moved to the same parish as where they were
originally convicted (S = 0). As revealed in Table 1, this situation certainly does not
hold—i.e., 53.2 percent of parolees released post-Katrina (Z = 1) moved to the same
parish (S = 0), and 26.1 percent of parolees from the two pre-Katrina cohorts (Z = 0)
moved to a different parish (S = 1). Thus, while it is adequate to assume that the
assignment to treatment is ignorable (i.e., conditional on the instrument Z and the other
covariates from equation (2), assignment to control and treatment groups is random), a
consequence of noncompliance is that the receipt of treatment is nonignorable (Angrist et
al. 1996). If this is the case, simply computing the difference between the pre-Katrina and
post-Katrina cohorts on recidivism will not provide an unbiased estimate of the average
causal effect of migrating to a different parish on recidivism. Yet, through the use of IV
methods, I am able to compute an unbiased estimate of the effect of migrating to a
34
different parish on re-incarceration for those parolees who otherwise would not have
moved had it not been for Hurricane Katrina (i.e., an estimate of LATE).12 Results
reported to follow represent the LATE estimate.
Table 4 presents the IV probit results of re-incarceration. With the exception of
the coefficient for age at release, the associations between individual-level covariates and
re-incarceration are similar to findings presented in Table 3. The key parameter of
interest is that for Different Parish as Conviction, which is a binary measure indicating
whether a given parolee resides in a different parish upon release relative to where s/he
was convicted prior to incarceration. Results show that those individuals who migrated to
a different parish were significantly and substantially less likely to be re-incarcerated
within one year of release. With respect to the marginal effect, the probability of reincarceration within one year of prison release is 0.16 lower for parolees who did not
move back to the parish where they were originally convicted relative to parolees who
did (net of individual, contextual, and criminal justice system correlates). The 95 percent
confidence interval of the marginal effect ranges from -0.080 to -0.232. The predicted
probability of re-incarceration for male parolees who returned to same parish as where
they were convicted is 0.244. In contrast, the predicted probability for males released to a
different parish is 0.084. As noted, these inferences pertain to the local average treatment
effect for compliers—i.e., the effect of migrating to a different parish for those parolees
12
In essence, through the IV method, I adjust the recidivism difference between the post-Katrina cohort
versus pre-Katrina cohorts for the fact that not all post-Katrina parolees moved to a different parish and not
all pre-Katrina parolees returned to the same parish. This adjustment yields the Local Average Treatment
Effect, which is computed as follows (Angrist et al. 1996):
{E[Yi | Zi =1] - {E[Yi | Zi = 0]} / {E[Si | Zi =1] - {E[Si | Zi = 0]} = α . With perfect treatment compliance, the
denominator equals one and the treatment effect α simply equals the Intention-to-Treat (ITT) effect.
35
who otherwise would have moved back to the same parish instead had Hurricane Katrina
not occurred.
[TABLE 4 ABOUT HERE]
SENSITIVITY ANALYSIS
To assess whether findings are robust to a different operationalization of recidivism, I reestimate equations (2) and (3) with a combined measure of re-incarceration and detainers.
Results are presented in Table 4. While there is some variation in the correlates of this
second measure of recidivism relative to the previous model, I still find that those
parolees who return to a different parish relative to where they were originally convicted
are significantly less likely to recidivate (re-incarcerated or detained) within one year of
release. The probability of re-incarceration or detention within one year of prison release
is 0.07 lower for parolees who did not move back to the parish where they were
originally convicted relative to parolees who did.
As an additional sensitivity analysis, I re-estimate equations (2) and (3) using an
alternative specification of instrumental variables. As described in the Introduction,
property damage from Hurricane Katrina varied by parish, with Jefferson Parish the least
severely impacted and St. Bernard and Plaquemines parishes most severely impacted.
Thus, as depicted in Figure 1, the likelihood of a parolee moving to a given parish postincarceration was influenced by the extent of housing destruction in the parish. Given that
both the time period of release from prison (pre- versus post-Katrina) and the geographic
variation in property damage affects whether a given parolee will return to the same
parish where s/he was originally convicted, I use an interaction between time period of
36
release and parish of conviction as an alternative specification of instrumental variables.13
With this specification, I assume that whether a parolee migrates to a different parish or
not depends upon whether s/he was released from prison before or after Hurricane
Katrina and where s/he was originally convicted. Additionally, I assume that the
combination of time of release and location of prior conviction only impacts reincarceration indirectly, through residential migration following incarceration.
Findings presented in Appendix Table A1 are consistent with inferences derived
in Table 4. Parolees who migrated to a different parish were significantly less likely to be
re-incarcerated, such that the marginal effect of re-incarceration within one year of prison
release is 0.14 lower for parolees who did not move back to the parish where they were
originally convicted relative to parolees who did.
In sum, findings suggest that moving away from former places of residence serves
some benefit for parolees in that a change of residence allows individuals to separate
from their peers and the temptations that contributed to their criminality in the first place.
This finding holds under alternative specifications of the dependent variable as well as
the instruments.
DISCUSSION
Selection bias is very much an issue with research on residential migration as well as
“neighborhood effects.” The innovation of this study has been to utilize a natural
experiment in order to investigate a theoretical question whose answer has largely eluded
13
In essence, I have five instruments: the first instrument (Post-Katrina release and Orleans conviction)
equals 1 for all parolees who were released following Katrina and who were originally convicted in
Orleans Parish, and equals 0 otherwise. I construct similar instruments for parolees originally convicted in
Jefferson, Plaquemines, St. Bernard, and St. Tammany parishes, respectively. All five instruments are
entered into equations (2) and (3) simultaneously.
37
researchers because of selection—just how consequential is a change of residence to
behavioral outcomes such as crime? In the absence of perfect information on why a given
individual moves to one place of residence versus another, it is problematic to answer
questions like this one. While tragic, the residential destruction and migration resulting
from Hurricane Katrina present a unique opportunity for understanding how place and
migration affect outcomes such as crime and recidivism.
In this study I have addressed whether recidivism declined post-Katrina and
whether separating individuals from their former criminogenic environment reduces their
likelihood of recidivism. Findings support both assertions. In particular, the marginal
effect of migrating to a different parish upon release from prison is a 0.16 decline in the
probability of re-incarceration.
Whether such a finding is internally valid is critical to consider. Recall my
discussion of the exclusion restriction, which states that the instrument and the outcome
variable are unrelated except indirectly through the effect of the instrument on the causal
treatment. In other words, the validity of my IV analysis rests on the assumption that any
effect of the time period of release (pre- versus post-Katrina) on the likelihood of reincarceration must be captured through the effect of time period on residential migration.
The instrument should not be related to any other factors besides residential migration
that are predictive of re-incarceration. Results presented in Table 4 reveal that none of the
contextual and criminal justice system covariates included in the analysis, which are
likely correlated with Hurricane Katrina, are predictive of incarceration. In other words,
my instrument has little partial effect on re-incarceration through these covariates
because none of these covariates are predictive of re-incarceration.
38
While I have utilized a variety of data in order to measure as many relevant
predictors of recidivism as possible, it is plausible that there are unobserved changes in
the post-Katrina context which are correlated with both my instrument and reincarceration, thus biasing my results. One such unobservable factor may be
neighborhood collective efficacy (Sampson, Raudenbush, and Earls 1997). If Hurricane
Katrina is correlated with temporal variation in collective efficacy and, in turn, collective
efficacy is related to re-incarceration, then I have violated the exclusion restriction and
may therefore have bias in the findings presented in Table 4. However, to my knowledge
there are no existing studies which investigate the association between collective efficacy
and recidivism. I therefore have little guidance as to whether collective efficacy is
correlated with recidivism at the individual-level. Still, it is instructive to point out that
while collective efficacy is highly predictive of neighborhood rates of crime (Morenoff,
Sampson, and Raudenbush 2001; Sampson et al. 1997), a number of recent studies reveal
that collective efficacy has a much weaker impact on arrest and violence at the
individual-level (Kirk 2008; Sampson, Morenoff and Raudenbush 2005). Moreover,
research generally supports the notion that geographic units, particularly at the parish or
county level, are far more internally heterogeneous than externally differentiable, such
that more within parish variability in criminal outcomes should be expected than between
parish variability (Cook, Shagle, and Degirmencioglu 1997). I use this discussion to
suggest that even if some unobserved contextual factor, such as collective efficacy, is
correlated with my instrument, it is very likely the case that it has at most a modest effect
on re-incarceration. In sum, I cannot rule out the possibility of violating the exclusion
restriction, but I suggest that even with such a violation the potential bias would not
39
wholly eliminate the sizable treatment effect observed in Table 4 (i.e., the 0.16 difference
in the probability of re-incarceration for those parolees who returned to the same parish
as where they were convicted relative to those who migrated to a different one).
The next step in this causal story is to further investigate why there is such a
powerful treatment effect from changing place of residence. Yet, even before the full
causal story bears out, findings established thus far with respect to residential migration
and recidivism have significant policy implications. If separating ex-prisoners from their
former residential environment actually benefits those prisoners (in terms of
reintegration) without sacrificing public safety, then a logical next step is to consider how
to disperse the population of ex-prisoners on a large scale. Research by the Urban
Institute suggests that substantial proportions of returning prisoners would welcome the
opportunity to move away from their former criminogenic environments (Visher and
Courtney 2006; Visher and Farrell 2005). For instance, in a sample of 400 recently
released males from the Illinois Department of Corrections, Visher and Farrell (2005)
find that 45 percent of returning prisoners explicitly expressed a desire to move to a
different neighborhood than where they lived prior to prison, with over half of this group
desiring to do so in order to avoid drugs and the other temptations and troubles of
familiar neighborhood settings. Yet many ex-prisoners still end up moving back to their
former counties and neighborhoods despite an expressed interest to avoid such places.
One prime reason is because of legal barriers. In most states, prisoners released to some
form of parole supervision are legally required to return to their county of last residence
(National Research Council 2007). Thus, criminal justice policies in most states are
designed to return prisoners to the same familiar surroundings where they got into trouble
40
with the law in the first place. In Louisiana, there are no such geographic restrictions.
Findings from Louisiana presented in this study suggest that allowing ex-prisoners to
move to different parishes or counties will reduce recidivism. Thus, it may be fruitful for
states to reconsider the residency restrictions imposed upon returning prisoners. Yet, as
noted, the sizable treatment effect of moving found in this study is applicable to just those
parolees who moved to a different parish post-Katrina who otherwise would have moved
back to the where they lived before. To fully explore the policy implications of this
finding, it is necessary to understand why prisoners return to their former parishes and
neighborhoods even in a state like Louisiana where there are no legal barriers preventing
parolees from moving to a different parish. We know from the same Urban Institute
research that another key reason why released prisoners move back to their former
neighborhoods is to be close to family and friends (Visher and Farrell 2005). Thus, to
capitalize on the apparent benefits of residential migration, it may be necessary to both
remove legal barriers for ex-prisoners to move as well as provide ex-prisoners and their
families with opportunities and incentives to move away from old neighborhoods. In
other words, forcing ex-prisoners to move away from their old neighborhoods is not
realistic or ethical. But providing opportunities for ex-prisoners to move away from old
neighborhoods is a policy prescription that may be worth pursuing.
In advance of enacting policy initiatives to promote residential change among the
parole population, more research needs to be undertaken to both validate the findings
demonstrated in this study and to uncover the mechanism underlying the treatment effect
(i.e., why it is that migration to new surroundings has such a consequential impact on the
likelihood of recidivism). With respect to the former, replicating the current study may
41
prove challenging given that the research design is based upon a natural experiment. That
said, similar investigations could be carried out in other geographic areas which have
been affected by hurricanes and other natural disasters. Another possibility for further
research is to contrast recidivism rates of states which require prisoners to return to their
county of last residence versus those states which do not. A third approach would be to
identify states which changed the residency restrictions imposed upon released prisoners
and then compare recidivism rates before and after such changes were made. Finally, as
Abbott (1997:89) notes, “what makes a turning point a turning point rather than a minor
ripple is the passage of sufficient time ‘on a new course’ such that it becomes clear that
direction has indeed been changed.” Thus, in order to more definitively conclude that
Hurricane Katrina and the resulting migration produced turning points in the life course
of crime for many former prisoners, it is necessary to follow-up the same cohorts of
individuals in future analyses to determine if the observed differences in recidivism
across cohorts remain after a sufficient passage of time (e.g., 3 to 5 years).
While more research on migration, place of residence, and recidivism is needed,
findings thus far suggest that successful prisoner reentry and reintegration depends upon
providing opportunities for prisoners to separate from their criminogenic past. Such a
finding provides initial support for the life-course propositions outlined previously. If
criminal peers and familiar criminogenic contexts causally influence criminal behavior,
then we should see that separating individuals from their peers and familiar contexts
should lead to a reduced likelihood of criminal behavior. That is precisely what I find. In
a practical sense, to lessen the likelihood of recidivism and to foster the path to
desistance, it is beneficial to separate ex-prisoners from their criminal past.
42
REFERENCES
Abbott, Andrew. 1997. “On the Concept of Turning Point.” Comparative Social
Research 16:85-105.
Angrist, Joshua D. 1990. “Lifetime Earnings and the Vietnam Era Draft Lottery:
Evidence from Social Security Administration Records.” American Economic
Review 88:450-77.
Angrist, Joshua D., Guido W. Imbens, and Donald B. Rubin. 1996. “Identification of
Causal Effects Using Instrumental Variables.” Journal of the American Statistical
Association 91:444-55.
Angrist, Joshua D. and Alan B. Krueger. 2001. “Instrumental Variable and the Search of
Identification: From Supply and Demand to Natural Experiments.” Journal of
Economic Perspectives 15: 69-85.
Brookings Institution, The. 2006. Katrina Index: Tracking Variables of Post-Katrina
Recovery. Washington: The Brookings Institution. Retrieved December 8, 2006
(http://www.brookings.edu/metro/pubs/200607_KatrinaIndex.pdf).
Brookings Institution, The and Greater New Orleans Community Data Center. 2007. The
New Orleans Index: Second Anniversary Special Edition. Washington: The
Brookings Institution. Retrieved August 21, 2007
(http://www.brookings.edu/reports/2007/~/media/Files/rc/reports/2007/08neworle
ansindex/200708_katrinaES.pdf).
Brown, Claude. 1965. Manchild in the Promised Land. New York: MacMillan.
Carter, Brian L. and Stephen T. Tiffany. 1999. “Meta-Analysis of Cue-Reactivity in
Addition Research.” Addiction 94:327-40.
Clear, Todd R., Elin Waring, and Kristen Scully. 2005. “Communities and Reentry:
Concentrated Reentry Recycling.” In Prisoner Reentry and Crime in America, J.
Travis and C. Visher (eds.), pp. 179-208. New York: Cambridge University Press.
Cook, Thomas D., Shobha C. Shagle, and Serdar M. Degirmencioglu. 1997. “Capturing
Social Process for Testing Mediational Models of Neighborhood Effects.” In
Neighborhood Poverty, Vol. II, eds. J. Brooks-Gunn, G.J. Duncan, and J.L. Aber,
pp. 94-119. New York: Russell Sage.
Dixon, Jo. 1995. “The Organizational Context of Criminal Sentencing.” American
Journal of Sociology 100:1157-98.
43
Duncan, Otis D. and Beverly Duncan. 1955. “A Methodological Analysis of Segregation
Indexes.” American Sociological Review 20:210-17.
Eisenstein, James, Roy Flemming, and Peter Nardulli. 1988. The Contours of Justice:
Communities and Their Courts. Boston: Little, Brown.
Elder, Glen H., Jr. 1974. Children of the Great Depression: Social Change in Life
Experience. Chicago: University of Chicago Press.
ESRI. 2006a. ESRI Demographic Update Methodology: 2006/2011. Redlands, CA:
ESRI. Retrieved October 24, 2007 (http://www.esri.com/library/whitepapers/pdfs/
demographic-update-methodology.pdf).
------. 2006b. ESRI Gulf Coast Updates Methodology: 2006/2011. Redlands, CA: ESRI.
Retrieved October 24, 2007 (http://www.esri.com/library/whitepapers/pdfs/gulfcoast-methodology.pdf).
------. 2007. Community Sourcebook of County Demographics, 19th Edition. Redlands,
CA: ESRI.
Feeley, Malcolm M. 1979. The Process is the Punishment. New York: Russell Sage.
Garrett, Brandon L. and Tania Tetlow. 2006. “Criminal Justice Collapse: The
Constitution after Hurricane Katrina.” Duke Law Journal 56:127-78.
Heumann, Milton. 1975. “A Note on Plea Bargaining and Caseload Pressure.” Law and
Society Review 9:515-28.
Hipp, John R. 2007. “Block, Tract, and Levels of Aggregation: Neighborhood Structure
and Crime and Disorder as a Case in Point.” American Sociological Review
72:659-80.
Johnson, Brian D. 2005. “Contextual Disparities in Guidelines Departures: Courtroom
Social Contexts, Guidelines Compliance, and Extralegal Disparities in Criminal
Sentencing.” Criminology 43:761-96.
Kirk, David S. 2008. “The Neighborhood Context of Racial and Ethnic Disparities in
Arrest.” Demography 45: 55-77.
La Vigne, Nancy G., Vera Kachnowski, Jeremy Travis, Rebecca Naser, and Christy
Visher. 2003. A Portrait of Prisoner Reentry in Maryland. Washington, DC:
Urban Institute.
La Vigne, Nancy G. and Cynthia A. Mamalian. 2004. A Portrait of Prisoner Reentry in
Georgia. Washington, DC: Urban Institute.
44
La Vigne, Nancy G., Cynthia A. Mamalian, Jeremy Travis, and Christy Visher. 2003. A
Portrait of Prisoner Reentry in Illinois. Washington, DC: Urban Institute.
La Vigne, Nancy G. and Gillian L. Thomson. 2003. A Portrait of Prisoner Reentry in
Ohio. Washington, DC: Urban Institute.
Langan, Patrick A. and David J. Levin. 2002. Recidivism of Prisoners Released in 1994.
Washington, DC: Bureau of Justice Statistics.
Laub, John H. and Robert J. Sampson. 2001. “Understanding Desistance from Crime.”
In Crime and Justice: A Review of Research, Volume 28, M. Tonry (ed.), pp. 169. Chicago: University of Chicago Press.
------. 2003. Shared Beginnings, Divergent Lives: Delinquent Boys to Age 70. Cambridge,
MA: Harvard University Press.
Liu, Amy, Bruce Katz, Matt Fellowes, and Mia Mabanta. 2005. Housing Families
Displaced by Katrina: A Review of the Federal Response to Date. Washington,
DC: Brookings Institution.
Logan, John R. 2006. The Impact of Katrina: Race and Class in Storm-Damaged
Neighborhoods. Providence, RI: Brown University, Spatial Structures in the
Social Sciences. Retrieved December 3, 2007
(http://www.s4.brown.edu/katrina/report.pdf).
Louisiana Public Health Institute. 2006. 2006 Louisiana Health and Population Survey:
Orleans Parish. New Orleans: Louisiana Public Health Institute. Retrieved
December 20, 2006
(http://popest.org/popestla2006/files/Popest_Orleans_SurveyReport.pdf).
Lynch, James P. and William J. Sabol. 2001. Prisoner Reentry in Perspective.
Washington, DC: Urban Institute.
Morenoff, Jeffrey D., Robert J. Sampson, and Stephen W. Raudenbush. 2001.
“Neighborhood Inequality, Collective Efficacy, and the Spatial Dynamics of
Homicide.” Criminology 39:517-60.
Mumola, Christopher J. and Jennifer C. Karberg. 2006. Drug Use and Dependence, State
and Federal Prisoners, 2004. Bureau of Justice Statistics Special Report.
Washington, DC: U.S. Department of Justice.
Nardulli, Peter F. 1979. “The Caseload Controversy and the Study of Criminal Courts.”
Journal of Criminal Law and Criminology 70:89-101.
National Research Council. 2007. Parole, Desistance from Crime, and Community
Integration. Committee on Community Supervision and Desistance from Crime.
45
Committee on Law and Justice, Division of Behavioral and Social Sciences and
Education. Washington, DC: The National Academies Press.
Osborn, S.G. 1980. “Moving Home, Leaving London and Delinquent Trends.” British
Journal of Criminology 20:54-61.
Pager, Devah. 2003. “The Mark of a Criminal Record.” American Journal of Sociology
108:937-75.
Roman, Caterina Gouvis, Seri Irazola, and Jenny W.L. Osborne. 2007. After Katrina:
Washed Away? Justice in New Orleans. Washington, DC: Urban Institute.
Rosenfeld, Richard, Joel Wallman, and Robert Fornango. 2005. “The Contribution of ExPrisoners to Crime Rates.” In Prisoner Reentry and Crime in America, J. Travis
and C. Visher (eds.), pp. 80-104. New York: Cambridge University Press.
Sabol, William J. and Paige M. Harrison. 2007. Prison and Jail Inmates at Midyear 2006.
Bureau of Justice Statistics Bulletin. Washington, DC: U.S. Department of
Justice.
Sampson, Robert J. and John H. Laub. 1993. Crime in the Making: Pathways and
Turning Points through Life. Cambridge, MA: Harvard University Press.
Sampson, Robert J., Jeffrey D. Morenoff, and Stephen W. Raudenbush. 2005. “Social
Anatomy of Racial and Ethnic Disparities in Violence.” American Journal of
Public Health 95:224-32.
Sampson, Robert J., Stephen W. Raudenbush, and Felton Earls. 1997. Neighborhoods
and Violent Crime: A Multilevel Study of Collective Efficacy. Science 227:91824.
Shaw, Clifford and Henry D. McKay. 1931. Report on the Causes of Crime, Volume II.
Washington, DC: U.S. Government Printing Office.
Sutherland, Edwin H. 1947. Principles of Criminology, 4th Edition. Philadelphia, PA: J.B.
Lippincott.
Turner, Susan and Joan Petersilia. 1992. “Focusing on High-Risk Parolees: An
Experiment to Reduce Commitments to the Texas Department of Corrections.”
Journal of Research in Crime and Delinquency 29:34-61.
Ulmer, Jeffery and Brian D. Johnson. 2004. “Sentencing in Context: A Multilevel
Analysis.” Criminology 42:137-75.
46
U.S. Department of Housing and Urban Development. 2006. Current Housing Unit
Damage Estimates: Hurricanes Katrina, Rita, and Wilma. Washington, DC: U.S.
Department of Housing and Urban Development. Retrieved December 8, 2006
(http://www.huduser.org/publications/destech/GulfCoast_HsngDmgEst.html).
Visher, Christy and Shannon M.E. Courtney. 2006. Cleveland Prisoners’ Experiences
Returning Home. Washington, DC: Urban Institute.
Visher, Christy and Jill Farrell. 2005. Chicago Communities and Prisoner Reentry.
Washington, DC: Urban Institute.
Warr, Mark. 1993. “Age, Peers, and Delinquency.” Criminology 31:17-40.
------. 1998. “Life-Course Transitions and Desistance from Crime.” Criminology 36:183216.
------. 2002. Companions in Crime: The Social Aspects of Criminal Conduct. Cambridge,
UK: Cambridge University Press.
Wikler, Abraham. 1948. “Recent Progress in Research on the Neurophysiologic Basis of
Morphine Addiction.” American Journal of Psychiatry 15:329-38.
Watson, Jamie, Amy L. Solomon, Nancy G. La Vigne, Jeremy Travis, Meagan Funches,
and Barbara Parthasarathy. 2004. A Portrait of Prisoner Reentry in Texas.
Washington, DC: Urban Institute.
47
Table 1. Comparison of Post-Incarceration Place of Residence
Post-Incarceration Residence
Same Parish
Different Parish
2001-2002
73.9%
26.1%
2003-2004
73.9%
26.1%
2005-2006
53.2%
46.8%
Cohort
Notes: "Same" means the proportion of parolees in a
given cohort who moved to the same parish following
incarceration as where they were originally convicted.
"Different" refers to the converse. Chi-Square =
127.405, p < 0.001.
48
Table 2. Percent of Ex-Prisoners by Category: Social and Economic Context
2001-2002
2003-2004
2005-2006
Chi-Square
P value
Dissimilarity
0 to 25
26 to 50
51 to 100
14.7%
79.9%
5.4%
12.4%
81.6%
6.0%
10.1%
78.6%
11.3%
37.432
<0.001
Average
Household
Income
< $25,000
$25 - $50,000
> $50,000
28.6%
68.5%
2.9%
26.7%
71.1%
2.2%
26.6%
70.7%
2.7%
2.762
0.598
Unemployment
Rate
0 to 5.0
5.1 to 10.0
10.1+
33.8%
66.2%
0.0%
41.6%
58.4%
0.0%
10.7%
53.7%
35.6%
1,032.155
<0.001
Average
Weekly Wage
< $500
$500 - $750
> $750
11.1%
88.5%
0.4%
7.5%
91.3%
1.2%
8.2%
70.9%
20.9%
452.056
<0.001
Fair Market
Rent
< $500
$500 - $750
> $750
7.9%
92.1%
0.0%
5.5%
94.5%
0.0%
18.2%
12.5%
69.3%
2,337.250
<0.001
Note: Average Household Income, Average Weekly Wage, and Fair Market Rent all adjusted to 2000 dollars.
49
Table 3. Intention-to-Treat Probit Estimates of Re-Incarceration
Re-Incarceration
Coef.
Post-Katrina Release
Robust
Std. Err.
-0.388
(0.090) ***
Individual-Level
Black
Male
Married
Age at Release
Time Served
0.072
0.222
-0.224
-0.008
-0.038
(0.048)
(0.068) ***
(0.073) **
(0.004)
(0.013) **
Context and Crim. Justice System
Unemployment Rate
Avg. Weekly Wage
Avg. Household Income (Sqrt)
Dissimilarity
Fair Market Rent
Ave. Parole Contacts
Judge Caseloads
UCR Arrests per Crime (Parish)
-0.061
-0.003
-0.004
-0.040
0.011
-0.015
0.000
-0.378
(0.086)
(0.005)
(0.005)
(0.036)
(0.004) ***
(0.012)
(0.001)
(0.614)
-0.696
(0.430)
Intercept
Notes: * p<=0.05
** p<=0.01 *** p<=0.001
Coefficients and standard errors for all Context and Criminal
Justice System measures except UCR Arrests per Crime are
multiplied by 10. Significance tests are calculated from robust
standard errors.
50
Table 4. Instrumental Variable Probit Estimates of Re-Incarceration
Re-Incarceration
Coef.
Different Parish as Conviction
Robust
Std. Err.
Re-Incarceration/Detainer
Coef.
Robust
Std. Err.
-0.685 (0.194) ***
-0.195 (0.093) *
Individual-Level
Black
Male
Married
Age at Release
Time Served
0.043
0.208
-0.191
-0.008
-0.035
(0.084)
(0.069)
(0.064)
(0.004)
(0.014)
0.107
-0.126
0.290
-0.013
-0.011
(0.074)
(0.078) ***
(0.073)
(0.003) ***
(0.007)
Context and Crim. Justice System
Unemployment Rate
Avg. Weekly Wage
Avg. Household Income (Sqrt)
Dissimilarity
Fair Market Rent
Ave. Parole Contacts
Judge Caseloads
UCR Arrests per Crime (Parish)
-0.014
0.001
0.007
-0.003
-0.009
-0.008
0.000
-0.481
(0.088)
(0.005)
(0.006)
(0.045)
(0.005)
(0.007)
(0.001)
(0.633)
-0.197
0.006
-0.006
-0.005
0.000
0.010
0.000
-0.141
(0.073)
(0.007)
(0.010)
(0.037)
(0.004)
(0.008)
(0.001)
(0.170)
Intercept
0.083 (0.379)
**
**
*
*
-0.430 (0.386)
Notes: * p<=0.05 ** p<=0.01 *** p<=0.001
The instrument Zi is a binary indicator of the release period (pre-hurricane versus posthurricane). Coefficients and standard errors for all Context and Criminal Justice System
measures except UCR Arrests per Crime are multipled by 10. Significance tests are
calculated from robust standard errors.
51
Appendix Table A1. Instrumental Variable Probit Estimates of Re-Incarceration
Re-Incarceration
Coef.
Different Parish as Conviction
Robust
Std. Err.
-0.619
(0.244) *
Individual-Level
Black
Male
Married
Age at Release
Time Served
0.047
0.210
-0.194
-0.008
-0.036
(0.089)
(0.072)
(0.059)
(0.004)
(0.013)
Context and Crim. Justice System
Unemployment Rate
Avg. Weekly Wage
Avg. Household Income (Sqrt)
Dissimilarity
Fair Market Rent
Ave. Parole Contacts
Judge Caseloads
UCR Arrests per Crime (Parish)
-0.042
0.000
0.007
-0.007
-0.007
-0.008
0.000
-0.519
(0.104)
(0.005)
(0.006)
(0.045)
(0.006)
(0.006)
(0.001)
(0.672)
0.037
(0.348)
Intercept
**
***
*
**
Notes: * p<=0.05 ** p<=0.01 *** p<=0.001
Equations (2) and (3) estimated with five instruments
representing the interaction between release period (prehurricane versus post-hurricane) and parish of conviction
(Orleans, Jefferson, Plaquemines, St. Bernard, and St.
Tammany parishes). Coefficients and standard errors for all
Context and Criminal Justice System measures except UCR
Arrests per Crime are multipled by 10. Significance tests are
calculated from robust standard errors.
52
Figure 1. Parish of Release
For Prisoners Committed from 5-Parish Area
50%
45%
% of All Releases to Parish
40%
35%
30%
25%
20%
15%
10%
5%
0%
Orleans
Jefferson
2001-2002
St. Tammany
St. Bernard
2003-2004
Plaquemines
East Baton
Rouge
Other
2005-2006
53
Figure 2. One-Year Re-Incarceration Rate
Prisoners Released to the Same Parish vs. a Different Parish
25%
22.3%
19.2%
20%
16.9%
15.0%
15%
13.9%
10.8%
10%
5%
0%
2001-2002
2003-2004
Same Parish
Different Parish
2005-2006